Heather – A Scottish Synthesised Voice at Last
Heather is a new synthesised voice from Cereproc. You can listen to her here;
http://www.cereproc.com
The best news is that that the CALL Centre, Edinburgh University have prompted the Scottish Government to licence the Heather voice for use in Scottish schools.
What does this mean?
Any software that can make use of a SAPI 5 voice will be able to speak in Heather’s voice. Clicker and Textease, software used mostly in primary schools, should be able to speak with this voice. It should work with the newly available WordTalk and Penfriend. The CALL Centre are currently testing the voice with these and other pieces of software. Heather should work with ‘Browse Aloud’ software which will read PDF files, including the digital exam papers from the SQA.
Heather should also work with software that turns text into MP3 audio files that pupils can listen to on an MP3 player. These can be low cost or cost free software such as the new version of Wordtalk (beta), or more sophisticated packages like Dolphin Producer that turns MS Word documents into DAISY format talking books.
Will it make a great difference?
It’s hard to tell. Whenever I demonstrate text-to-speech software there is often someone who complains about the quality of the speech. I imagine that they are complaining about the lack of emphasis that human speakers give to important words in a sentence. Text to Speech is not as easy to listen to as speech recorded from a skilful reader. It needs to be recognised that it is a different medium, which requires more concentration from the listener. However the difference in the speed of production is great. It can take about eight hours to produce a good recording of an hour of a person reading a text. If you have the text as a text file on a computer, you can let the computer ‘read’ the text into an audio file while you do something else.
Does the fact that it is Scottish matter? Will this boost the use of text to speech to support pupils in Scottish schools?
Most people say that it does. However I have heard visually impaired people listening to HAL, a fairly old, robotic sounding voice, with the reading speed turned up to a rate that I found very difficult to follow. Perhaps it’s your level of needs that makes you persevere.
It seems odd to me that Professor Stephen Hawking has the voice that sounds like a DECtalk speech synthesiser of the late 1980s, yet people complain about the SAPI 5 voices on modern computers being robotic or American. When we listen to a voice on the radio, we can pick up almost immediately the part of the country the speaker is from and the speaker’s level of education. We tend to react to speakers on class basis. If you need any proof of this, go the the AT&T Natural Voices. You can type in some text and hear them spoken by Charles or Camilla. ( Sorry that should be Audrey.)
http://www.naturalvoices.att.com/demos/
Someone has played a cruel joke on the developers by suggesting that UK users would appreciate having text read out by upper class twits. Do waste some time on this site, creating conversations between these two characters.
Anyway, Heather will, I hope, be available to be installed on computers in Scottish schools, at no additional cost this summer. Then we’ll find out the answer to these questions.
